Event-based egocentric dynamic networks are an important class of networks widely seen in many domains. In this paper, we present a visual analytics approach for these networks by combining data-driven network simplifications with a novel visualization design - EgoNetCloud. In particular, an integrated data processing pipeline is proposed to prune, compress and filter the networks into smaller but salient abstractions. To accommodate the simplified network into the visual design, we introduce a constrained graph layout algorithm on the dynamic network. Through a real-life case study as well as conversations with the domain expert, we demonstrate the effectiveness of the EgoNetCloud design and system in completing analysis tasks on event-based dynamic networks. The user study comparing EgoNetCloud with a working system on academic search confirms the effectiveness and convenience of our visual analytics based approach.
展开▼
